Comparison of signal drop on T2-weighted and post-contrast T1-weighted Dixon MRI in spinal bone marrow lesions.

Objective

To compare signal drop percentages between in-phase and out-of-phase images on T2-weighted and post-contrast T1-weighted Dixon MRI in spinal bone marrow lesions and evaluate their diagnostic performance in differentiating benign from malignant lesions.

Materials And Methods

This retrospective study included patients who underwent post-contrast spine MRI with 3.0 T Dixon sequences between January 2023 and January 2024. Signal drop percentages ([in-phase - out-of-phase] / in-phase) were calculated for T2-weighted Dixon (T2 %drop) and post-contrast T1-weighted Dixon images (GdT1 %drop). T2 %drop and GdT1 %drop were compared using the Wilcoxon signed-rank test. Diagnostic performance was assessed using receiver-operating characteristic analysis with area under the curve (AUC) calculation. Optimal cutoff values were determined to achieve 100 % sensitivity for malignancy while maximizing specificity. Specificities were compared using the McNemar test.

Results

Among 78 patients (median age, 63.5 years [interquartile range (IQR), 54.3-70.8]; 37 men, 41 women; 41 benign, 34 malignant, and 3 indeterminate lesions), median T2 %drop and GdT1 %drop were 29.2 % (IQR, 20.3-46.0) and 21.4 % (IQR, 15.4-28.6), respectively. GdT1 %drop was significantly lower than T2 %drop (P < 0.001). The AUC of GdT1 %drop (0.84; 95 % confidence interval [CI], 0.73-0.92) was numerically lower than that of T2 %drop (0.91; 95 % CI, 0.81-0.97), although the difference was not significant (P = 0.132). The specificity of GdT1 %drop was significantly lower than that of T2 %drop at the optimal cutoff values of 27.9 % and 29.0 %, respectively (39.0 % [95 % CI, 24.2-55.5] vs. 80.5 % [95 % CI, 65.1-91.2]; P < 0.001).

Conclusion

GdT1 %drop was significantly lower than T2 %drop with substantially lower specificity, warranting cautious interpretation to avoid misclassification.